readable AI benchmarks

Anthropic

Claude 4.5 Haiku (Non-reasoning)

non-reasoning · closed weights · Oct 15, 2025

Quality Score27.1
Value Score19.9
Reliability46.2
Cache Discount90%

Model specification

Reasoning
non-reasoning
Input modalities
text, image
Output modalities
text
Context window
200k tokens
Weights
closed weights

Token prices USD per 1M tokens

Input
$1.00
Output
$5.00
Cached Input
$0.100
Cached Output
$1.25
Cost per task
not available

Capability

Intelligence
24.1
Coding
not available
Agentic
not available
Omniscience
-7.6
Correct
14.4%
Blended price
$0.770

Answer outcomes

Correct 14.4%Incorrect 22.0%Abstained 63.6%

Artificial Analysis benchmarks

SciCode
34%
Humanity’s Last Exam
4%
GPQA Diamond
65%
CritPt
0%
AA-Omniscience
46%
AA-LCR
48%

Similar models